Researchers who proposed a connection between urinary levels of bisphenol A (BPA) in humans and the incidence of type 2 diabetes and heart disease did not adequately address the potential for a false positive result, said two statisticians in a Feb. 18, 2009 letter published in the Journal of the American Medical Association (JAMA). In the letter, Dr. S. Stanley Young, Assistant Director of Bioinformatics at the National Institute of Statistical Sciences, and Ming Yu, University of British Columbia, pointed out statistical limitations in the study, published Sept. 16, 2009 by Dr. Ian A. Lang and colleagues in JAMA. (The National Institute of Statistical Sciences (NISS) was established in 1991 by the national statistical societies and the Research Triangle universities and organizations, which includes Duke University and the University of North Carolina at Chapel Hill.)
The study by Lang and his colleagues used the 2003-2004 CDC National Health and Nutrition Examination Survey, which measures 275 environmental chemicals and a wide range of health outcomes. Given the number possible health outcomes—including combinations—that could be potentially associated with any of the 275 chemicals, along with multiple confounders and statistical models, there could be as many as approximately nine million statistical models available to analyze the data, Young and Yu reported. Since the Lang et al study focused on only one chemical and 16 health outcomes, Young and Yu concluded that the study's findings could likely be the result of chance rather than representing real health concerns. (The National Institute of Statistical Sciences issued a press release on the JAMA letter.)